METABOLOMICS ANALYSIS OF BLOOD AND URINE TO IDENTIF Y

ALCOHOL-DEPENDENCE BIOMARKERS

by

HAMZA MOHAMED AMIN MOSTAFA

Thesis submitted in fulfillment of the requirements

For the Degree of Master of Science

July2015

ÉΟó¡ Î0 «! $# Ç≈uΗ÷q §�9$# ÉΟŠÏm§�9 $#

* y7 tΡθ è=t↔ó¡o„ Ç∅tã Ì� ôϑy‚ ø9 $# Î�Å£÷�yϑø9 $#uρ ( ö≅è% !$ yϑÎγŠÏù ÖΝøOÎ) ×�� Î7Ÿ2 ßìÏ�≈ oΨtΒ uρ Ĩ$ ¨Ζ=Ï9 !$ yϑßγ ßϑøOÎ)uρ

ç�t9 ò2r& ÏΒ $ yϑÎγ Ïèø�̄Ρ 3 š�tΡθ è=t↔ó¡o„ uρ #sŒ$tΒ tβθà)Ï�ΖムÈ≅ è% uθø�yè ø9 $# 3 š�Ï9≡ x‹ x. ß Îit7 ムª!$# ãΝä3s9

ÏM≈tƒ Fψ$# öΝà6 ¯=yès9 tβρã� ©3x�tF s? ∩⊄⊇∪

219آ�� : رة ا����ة

(They ask you about wine and gambling. Say, "In them is great sin and [yet,

some] benefit for people. But their sin is greater than their benefit." And they

ask you what they should spend. Say, "The excess [beyond needs]." Thus

Allah makes clear to you the verses [of revelation] that you might give

thought).

Quran, Surah 2 (Al-Baqara) : 219

DEDICATION

I would like to dedicate this work to my parents and my siblings Arwa,

Moaaz, Anas, Baraa and Somaya, for their unconditional encouragement and support

throughout my Master.

ii

ACKNOWLEDGMENT

First and foremost, all the praises and the thanks are due to Allah the

Almighty, the God of the world and hereafter, for his showers of blessings,

protecting and guiding me throughout my study, and to the Prophet Muhammad

(prayers and peace be upon him), the beacon of knowledge. I could never have

accomplished this study without the faith I have in the Allah Almighty.

I would like to express my deep and sincere gratitude to my supervisor, Dr.

Baharudin Ibrahim for the constant support, untiring help, guidance and motivation

throughout my Master study. His vision and dynamism have inspired and motivated

me. It was a great honor to study and work under his supervision. I would also like to

thank him for his friendship and great sense of humor. He has given me enough

freedom duringmy research, and he has always been gracious to me. Also, I would

like to thank my co-supervisor, Dr. Vikneswaran a/l Murugaiyah for his guidance

and help. I appreciate his help and contribution tothe methodology of the research

and to present the research works as clearly as possible.

I would like to thank Dr. Nor Hayati Arif, the head of the Psychiatry

Department in Hospital Pulau Pinang for giving me the opportunity to do the

research in her department and providing the invaluable guidance throughout this

research. Also, I would like to thank staff nurse Hanisah and sister Subaidah and all

the staff nurses of the Psychiatry Department for helping me to find volunteers for

this study and to take blood samples from them. Above all, none of the work would

have been possible without the dedication of the AD patients, social drinkers and

controls who volunteered in the study.

iii

I would like to thank Dr. Teh Chin-Hoe from Bruker (Malaysia) Sdn Bhd for

his guidance and help in using NMR spectrometer and identifying the biomarkers.

My deepest thanks to my fellow laboratory mates in the clinical pharmacy lab

of our school, and laboratory assistants for assisting and supporting me. My research

could not have been completed without their help and contributions. I would like to

extend my heartfelt gratitude to the laboratory assistants of the pharmaceutical

chemistry lab of our school for assisting me in various laboratory activities and

providing valuable suggestions for my research.

I am deeply grateful to my parents for their love, prayers, sacrifices, care

andtheir moral and financial help for my education and preparing me for my future. I

have always felt that it is impossible to describe their support in words. I would also

like to thank my beloved elder sister Arwa, for always being by my side with her

tremendous support throughout my study journey. A special thank to my younger

sister Somaya, the star who brightens up my darkest hour and outshines everyone

that surrounds her. I acknowledge her efforts in adjusting the figures and charts in

my thesis. I would alsothank my elder brothers Moaaz, Anas and Baraa for their

support and encouragement throughout my study.

Special thanks to my friends from my school and other USM-schools such as

school of Chemistry and school of Biology for the assistance and valuable

discussions throughout my study. I am extending my thanks to myMalaysian and

international friends from different countries and to my coaches and teammates in

USM football team who have been the best family during my stay in Penang. I thank

you all for the moments that we spent together, for the friendships and comforting

words through rough times in my study.

iv

TABLE OF CONTENTS

Page

DEDICATION

ACKNOWLEDGMENT .............................................................................................. ii

TABLE OF CONTENTS.............................................................................................iv

LIST OF TABLES ...................................................................................................... ix

LIST OF FIGURES ..................................................................................................... x

LIST OF SYMBOLS AND ABBREVIATIONS ...................................................... xii

LIST OF PUBLICATIONS AND CONFERENCES ................................................ xv

ABSTRAK ............................................................................................................... xvii

ABSTRACT ............................................................................................................... xx

CHAPTER 1 ................................................................................................................ 1

INTRODUCTION AND LITERATURE REVIEW .................................................... 1

1.0 Background ................................................................................................... 1

1.1 The global burden of alcohol drinking .......................................................... 1

1.2 Alcohol consumption in Malaysia ................................................................. 2

1.3 Organ damage induced by alcohol consumption........................................... 2

1.4 Alcohol metabolism in human ..................................................................... 4

1.5 Alcohol-dependence (AD) ............................................................................ 4

1.5.1 Current diagnostic methods of alcohol-dependence (AD) ..................... 6

1.6 Systems Biology .......................................................................................... 12

v

1.6.1 Disciplines of Systems Biology ........................................................... 13

1.7 Metabolomics .............................................................................................. 14

1.8 Application of metabolomics in disease diagnosis ...................................... 16

1.8.1 Application of metabolomics techniques to identify novel-biomarkers

of diseases ........................................................................................................... 16

1.8.2 Using metabolomics to investigate alcohol consumption .................... 20

1.8.3 Using metabolomics to investigate alcohol consumption in animal

models 20

1.8.4 Using metabolomics to investigate alcohol consumption in humans .. 21

1.9 Application of metabolomics to find novel biomarkers of AD in blood and

urine samples .......................................................................................................... 22

1.10 Using 1H-NMR spectroscopic instrument in metabolomics ....................... 23

1.11 Data acquisition, mining and interpretation in metabolomics ..................... 25

1.11.1 Statistical approaches in plasma and urine metabolomics ............... 25

1.12 Rationale of the study .............................................................................. 27

1.13 Study hypothesis .......................................................................................... 28

1.14 Aims and objectives .................................................................................... 28

1.14.1 General objective ................................................................................. 28

1.14.2 Specific objectives ............................................................................... 28

CHAPTER 2 .............................................................................................................. 29

METHODOLOGY ..................................................................................................... 29

2.1 Study design and sample size ...................................................................... 29

vi

2.2 Study Participants ........................................................................................ 29

2.2.1 Recruitment of Alcohol-dependent (AD) patients ............................... 29

2.2.2 Recruitment of social drinkers (alcohol drinkers but not dependent) .. 30

2.2.3 Recruitment of Controls ....................................................................... 31

2.3 Sample collection and storage ..................................................................... 32

2.3.1 Sample collection and transfer to the lab ............................................. 32

2.3.2 Samples storage .................................................................................... 35

2.4 Samples preparation for metabolomics analysis ......................................... 37

2.4.1 Phosphate buffer preparation ............................................................... 37

2.4.2 Preparing the samples for 1HNMR analysis......................................... 37

2.5 Nuclear magnetic resonance (NMR) spectroscopy ..................................... 39

2.6 Data acquisition and pre-processing ............................................................ 39

2.7 Statistical pre-processing ............................................................................. 42

2.7.1 Binning ................................................................................................. 42

2.7.2 Normalization ....................................................................................... 42

2.7.3 Scaling .................................................................................................. 43

2.8 Data analysis ................................................................................................ 47

2.9 Metabolites Identification ............................................................................ 50

2.10 Reproducibility ............................................................................................ 54

2.11 Ethical Approval to conduct the study ........................................................ 55

CHAPTER 3 .............................................................................................................. 56

RESULTS .................................................................................................................. 56

vii

3.1 Demographic data ........................................................................................ 56

3.2 Identification of Ethanol peaks .................................................................... 58

3.3 Discriminating metabolic fingerprint in plasma .......................................... 61

3.3.1 Discriminating metabolic fingerprint using SIMCA............................ 61

3.3.2 Discriminating metabolic fingerprint in plasma using logistic

regression ............................................................................................................ 66

3.4 Discriminating metabolic fingerprint in urine ............................................. 72

3.4.1 Discriminating metabolic fingerprint using SIMCA............................ 72

3.4.2 Discriminating metabolic fingerprint using logistic regression ........... 79

3.5 Identification of biomarkers ........................................................................ 83

3.5.1 Plasma Biomarkers............................................................................... 83

3.5.2 Urine Biomarkers ................................................................................. 86

3.5.3 Confirmation of using acetic acid as a plasma biomarker for AD ....... 90

3.6 Reproducibility ............................................................................................ 92

3.6.1 Reproducibility of the results in plasma............................................... 93

3.6.2 Reproducibility of the results in urine .................................................. 94

CHAPTER 4 .............................................................................................................. 97

DISCUSSION ............................................................................................................ 97

4.0 Introduction ................................................................................................. 97

4.1 Demographic data ........................................................................................ 98

4.2 Exclusion of Ethanol peaks ......................................................................... 99

4.3 Discriminating metabolic fingerprint ........................................................ 100

viii

4.3.1 Discriminating metabolic fingerprint in plasma using SIMCA ......... 100

4.3.2 Discriminating metabolic fingerprint in plasma using logistic

regression .......................................................................................................... 101

4.3.3 Discriminating metabolic fingerprint in urine using SIMCA ............ 102

4.3.4 Discriminating metabolic fingerprint in urine using logistic regression

102

4.4 Identification of biomarkers ...................................................................... 103

4.4.1 Plasma ................................................................................................ 103

4.4.2 Urine ................................................................................................... 104

4.5 Reproducibility .......................................................................................... 106

CHAPTER 5 ............................................................................................................ 107

CONCLUSION ........................................................................................................ 107

5.1 Objectives accomplishment ....................................................................... 107

5.2 Limitations and recommendation for further studies ................................ 107

REFERENCES ......................................................................................................... 110

ix

LIST OF TABLES

Page

Table 1.1 Examples of metabolomics studies to diagnose diseases 19

Table 3.1 Frequencies and percentages of demographics/clinicaldata 57

Table 3.2 Sensitivity, specificity and accuracy of OPLS-DA model 64

Table 3.3 Chemical shifts of the bins with VIP value > 1 66

Table 3.4 Backward stepwise logistic regression model for AD vs 68 combined group

Table 3.5 Forward stepwise logistic regression model for AD vs 68 combined group

Table 3.6 Enter logistic regression model for AD vs combined groups 69

Table 3.7 Sensitivity, specificity and accuracy of the logistic regression 69 model

Table 3.8 Sensitivity, specificity and accuracy of OPLS-DA model 76

Table 3.9 Chemical shifts of the bins with VIP value> 1 78

Table 3.10 Rotated factor loading of pattern matrix 80

Table 3.11 Backward stepwise logistic regression model for AD vs 81 combined group

Table 3.12 Forward stepwise logistic regression model forAD vs 81 combined group

Table 3.13 Sensitivity, specificity and accuracy of thelogistic regression 82 model

Table 3.14 Molecular formula and chemical structure of plasma biomarkers 86

Table 3.15 Molecular formula and chemical structure of urine biomarkers 89

Table 3.16 Mean AUC of acetic acid for AD and social drinkers 91

Table 3.17 ICCs of the plasma biomarkers 94

Table 3.18 ICCs of the urine biomarkers 96

x

LIST OF FIGURES

Page

Figure 1.1 Ethanol metabolism in human body 4

Figure 1.2 The role of systems' biology disciplines (omics) in disease 14 diagnosis, where metabolomics represents the ultimate biomarker for detection of the disease before the phenotypic manifestations

Figure 1.3 Literature search at Google scholar for publications of 24 (metabolomics + NMR) and (Metabonomics + NMR)

Figure 2.1 The general scheme of metabolomics analysis: collecting, 34 storing and analyzing the samples

Figure 2.2 Blood sampling, separation of plasma and storing of 36 plasma and urine

Figure 2.3 Preparation of NMR samples 38

Figure 2.4 Steps of 1HNMR experiment setups in 41

Figure 2.5 Importing plasma data from Topspin to AMIX 45

Figure 2.6 Importing urine data from Topspin to AMIX 46

Figure 2.7 Steps of statistical analysis for plasma 49

Figure 2.8 Steps of statistical analysis for urine 50

Figure 2.9 Chenomx database 52

Figure 2.10 Human Metabolome Database 53

Figure 2.11 Biological Magnetic Resonance Data Bank 54

Figure 3.1 Urine 1HNMR spectra of the social drinker volunteer 60 Figure 3.2 Outliers in score scatter plot for controls group 62 Figure 3.3 Outliers in score scatter plot for AD group 62

Figure 3.4 Outliers in score scatter plot for social drinkers group 63

Figure 3.5 PCA-X overall groups plot 63

Figure 3.6 OPLS-DA plot 64

xi

Figure 3.7 VIP plotof OPLS-DA model 65

Figure 3.8 ROC curve (AD group vs combined group) 70

Figure 3.9 Plasma 1HNMR spectra of the three groups (first biomarker) 71

Figure 3.10 Plasma 1HNMR spectra of the three groups (second biomarker) 72

Figure 3.11 Outliers in score scatter plot for controls group 74

Figure 3.12 Outliers in score scatter plot for AD group 74

Figure 3.13 Outliers in score scatter plot for social drinkers group 75

Figure 3.14 PCA-X overall groups plot 75

Figure 3.15 OPLS-DA plot 76

Figure 3.16 VIP plotof OPLS-DA model 77

Figure 3.17 ROC curve (AD group vs combined group) 82

Figure 3.18 Example of a putative biomarker in urine 1HNMR spectra 83 of the three groups

Figure 3.19 Putative plasma biomarkers 84

Figure 3.20 Plasma1HNMR spectra of one alcohol-dependent subject with 85 theplasma 1HNMR spectra of the social drinker volunteer Figure 3.21 Putative urine biomarkers 88

Figure 3.22 Boxplot of acetic acid 91

Figure 3.23 Acetic acid peak in plasma 1HNMR spectra of the 92 social drinker volunteer

Figure 3.24 plasma 1HNMR spectra for one subject on two different days 93

Figure 3.25 Urine 1HNMR spectra for one subject on two different days 95

Figure 4.1 Ethanol metabolism and its interaction with NADH/NAD+ ratio 105

xii

LIST OF SYMBOLS AND ABBREVIATIONS

AD Alcohol-Dependence

ADH Alcohol dehydrogenase

ALDH Acetaldehyde dehydrogenase

ALT Alanine Aminotransferase

ANNs Artificial Neural Networks

AST Aspartate Aminotransferase

ASM Acid Sphingomyelinase

APTT Activated Partial Thromboplastin Time

AUC Area under the Curve

AUDIT Alcohol Use Disorders Identification Test

AUDIT-C Alcohol Use Disorders Identification Test-Consumption

AUROC Area under the receiver operating characteristic

B-BIOREFCODE Bruker Biofluid Reference Compound Database

BML Birmingham Metabolite Library

BMRB Biological Magnetic Resonance Data Bank

CDT Carbohydrate-Deficient Transferrin

COPD Chronic Obstructive Pulmonary Disease

CPMG Carr Purcell Meiboom Gill

CVD Cardiovascular Diseases

CYP2E1 Cytochrome P450

df Degrees of Freedom

DFA Discriminant Function Analysis

D2O Deuterium Oxide

DSM-IV Diagnostic and Statistical Manual of Mental Disorders-IV

xiii

ETG Ethyl Glucuronide

ETS Ethyl Sulfate

FDA Food and Drug Administration

g Gravity

g Gram

GGT Gamma Glutamyltransferase

HIV Human Immunodeficiency Virus

HMDB Human Metabolome Database

HPP Hospital Pulau Pinang

ICC Intra-Class Correlation Coefficient

Jres J-resolved

KMO Kaiser-Meyer-Olkin

LC Liquid Chromatography

LD Light Drinkers

MAST Michigan Alcoholism Screening Test

MCV Mean Corpuscular Volume

MHD Moderate to High Drinkers

ml milliliter

mM millimolar

MREC Medical Research and Ethics Committee

MS Mass Spectroscopy

NAD Nicotinamide Adenine Dinucleotide

Na2HPO4 Di-Sodium Hydrogen Phosphate

NaH2PO4 Sodium Dihydrogen Phosphate

NaN3 Sodium Azide

xiv

NIH National Institute of Health

NMR Nuclear Magnetic Resonance

NOESY Nuclear Overhauser Enhancement Spectroscopy

OPLS-DA Orthogonal Partial Least Square-Discriminated Analysis

PC Principal Component

PCA Principal Component Analysis

PEth Phosphatidylethanol

PLS Partial Least Square

ppm Parts Per Million

PT Prothrombin Time

SD Standard Deviation

SE Standard Error

SIMCA Soft Independent Modeling Of Class Analogies

TSP Sodium 3-(Trimethylsilyl)-Propionate-2,2,3,3-d4

UV Unit Variance

VIP Variable Influence on Projection

WHO World Health Organization

µL microliter

xv

LIST OF PUBLICATIONS AND CONFERENCES

Journal Paper /Abstract

Hamza Mohamed Amin Mostafa, Arwa Mohamed Amin Mostafa, Nor Hayati Arif,

Chin-Hoe Teh, Vikneswaran a/l Murugaiyah&Baharudin Ibrahim

(2014).Metabolomic analysis of blood andurine to identify alcohol-

dependencebiomarkers. The Medical Journal of Penang Hospital, Supplement 2014.

P.9.

Hamza Mostafa, Arwa M. Amin, Nor Hayati Arif, Chin-Hoe Teh, Vikneswaran a/l

Murugaiyah, Baharudin Ibrahim (2015). Nuclear magnetic resonance spectroscopy

based metabolomics to identify novel biomarkers of alcohol dependence. Asia-

Pacific Psychiatry journal.Submitted.

Hamza Mostafa, Arwa M. Amin, Nor Hayati Arif, Chin-Hoe Teh, Vikneswaran a/l

Murugaiyah, Baharudin Ibrahim (2015).Metabolomics Analysis of plasma and urine

to Identify Alcohol-Dependence biomarkers.(In processing of submission).

Conference Proceeding / Book of Abstract / Oral presentation /Poster

Presentation

Hamza Mohamed Amin Mostafa, Arwa Mohamed Amin Mostafa, Nor Hayati Arif,

Chin-Hoe Teh, Vikneswaran a/l Murugaiyah&Baharudin Ibrahim (2014).

Identification of Alcohol-Dependence Biomarkers in plasma by using Metabolomics

Analysis.The 14thasian conference on clinical pharmacy-Terengganu, Malaysia.

xvi

Hamza Mostafa, Arwa M. Amin, Nor Hayati Arif, Chin-Hoe Teh, Vikneswaran a/l

Murugaiyah, Baharudin Ibrahim (2015).Identification of Alcohol-Dependence

Biomarkers in urine by using Metabolomics Analysis.The 12th MPS Pharmacy

Scientific Conference-Kuala Lumpor, Malaysia.

xvii

ANALISIS METABOLOMIK DARAH DAN URIN UNTUK MENGENAL

PASTI PENANDA BIOLOGI KEBERGANTUNGAN ALKOHOL

ABSTRAK

Penyalahgunaan alkohol boleh membinasakan kesihatan masyarakat dan

mengakibatkan masalah sosial yang teruk.Kebergantungan alkohol merupakan fasa

penyalahgunaan alkohol sebagai akibat pengambilan alkohol dalam kuantiti yang

berlebihan dan sentiasa terdorong untuk mengambil alkohol secara

berterusan.Kaedah sedia ada untuk mengenalpasti kebergantungan alkohol adalah

melalui soal selidik dan beberapa penanda biologi.Malangnya, kedua-dua kaedah ini

mempunyai sensitiviti dan spesifisiti yang rendah.Metabolomik adalah lapangan

saintifik yang novel yang menyediakan kaedah novel untuk mengenalpasti

kebergantunagn alkohol dengan menggunakan teknik sensitif and spesifik seperti

resonans magnet nukleus (NMR).Tujuan kajian ini adalah mengenalpasti

pencapjarian metabolit (“metabotype”) yang dapat membezakan antara

kebergantunagn alkohol, peminum tetapi tidak bergantung pada alkohol dan individu

bebas alkohol dengan meggunakan kaedah metabolomik.Kajian ini turut dikaji dalam

kebolehulangan. Sampel darah dan urin dikumpul daripada 30 individu yang

bergantungan kepada alkohol (purata umur: 45.7), 54 peminum sosial (purata umur:

39.5) dan 60 kawalan (purata umur: 37.1). Plasma dipisahkan melalui pengemparan,

kemudian sampel plasma dan urin dicampur bersama larutan tampan fosfat yang

disediakan dengan melarutkan di-natrium hidrogen fosfat (Na2HPO4), natrium

dihidrogen fosfat(NaH2PO4), natrium 3-( trimetilsilil )- propionat -2,2,3,3-d4 (TSP),

natrium azida (NaN3) and deuterium oksida (D2O) seterusnya analisis melalui

spektroskopi NMR dijalankan. Data dianalisis menggunakan analisis multivarian

termasuk analisis komponen utama dan analisis perbezaan ortogon separa kuasa dua

xviii

terkecil (OPLS-DA) diikuiti regresilogistik univariate dan multivariat untuk

membangunkan model bagi pengenalpastian penanda biologi AD.Manakala, analisis

kebolehulangan menggunakan pekali kolerasi antara-kelas (ICC).

Kajian dalam plasma, model OPLS-DA mendedahkkan 39 pembolehubah

yang mempengaruhi unjuran(VIPs) secara signifikan yang dapat membezakan antara

kumpulan AD dengan peminum sosial dan kawalan. Kesensitifan, kespesifikan dan

kejituan model ini masing-masing adalah 64.29%, 98.17% dan 91.24%. Manakala

dalam urin, model OPLS-DA menunjukkan 59 VIPs yang membezakan secara

signifikan antara AD daripada peminum sosial dan kawalan. Kesensitifan,

kespesifikan dan kejituan model ini masing-masing adalah 86.21%, 97.25%

and94.93%. Analisis dalam plasma menggunakan regresi logistik univariat

mendapati 9 puncak yang berkaitan secara signifikan dengan AD dengan nilai p ≤

0.1. Seterusnya analisis dalam regresi logistik multivariat membuktikan 4 puncak

secara signikan berkait dengan AD dengan luas permukaan koefisi lingkungan

penerima (AUROC), 0.961 . Kesensitifan, kespesifikan dan kejituan model ini

masing-masing adalah 78.6%, 98.2% dan 94.2%. Manakala analisa dalam urin

menggunakan regresi logistik univariate, 30 puncak dikenalpasti mempunyai kaitan

signifikan dengan AD dengan nilai p ≤ 0.1. Analisis menggunakan PCA, mendapati

5 PCs mempunyai kaitan yang signikan dengan AD. Kemudian melalui analisis

regresi logistik multivariate, mendedahkan 2 PCs yang mempunyai 18 puncak

adalah berkait secara signifikan dengan AD dengan AUROC, 0.909. . Kesensitifan,

kespesifikan dan kejituan model ini masing-masing adalah 65.5%, 99.1%and92%.

xix

Melalui kajian ini, aplikasi teknik metabolomik menggunakan plasma dan

urin dapat membezakan antara individu kebergantungan alkohol , peminum sosial

dan kawalan. Metabolomik berdasarkan resonans magnetik nukleus dapat

mengenalpasti penanda biologi yang novel dalam plasma dan urin bagi diagnosis AD

untuk kegunaan masa hadapan.

xx

METABOLOMICS ANALYSIS OF BLOOD AND URINE TO IDENTIF Y

ALCOHOL-DEPENDENCE BIOMARKERS

ABSTRACT

Alcohol misuse is a ravaging public health and social problem. Alcohol-

dependence (AD) is a phase of alcohol misuse in which the drinker consumes

excessive amount of alcohol and have a continuous urge to consume alcohol. The

current methods of alcohol dependence diagnosis are questionnaires and some

biomarkers. However, both methods lack specificity and sensitivity. Metabolomics is

a novel scientific field which may provide a novel method for the diagnosis of

alcohol-dependence by using a sensitive and specific technique such as nuclear

magnetic resonance (NMR). Our aim was to identify the metabolic fingerprint

(metabotype) that is able to discriminate between AD, non-AD alcohol drinkers and

controls using metabolomics approach. The reproducibility of the outcomes was also

investigated. Blood and urine samples were collected from 30 alcohol-dependent

(mean age: 45.7), 54 social drinkers (mean age: 39.5) and 60 controls (mean age:

37.1). Plasma was separated by centrifugation, then both plasma and urine samples

were mixed with phosphate buffer which was prepared by dissolving di-sodium

hydrogen phosphate (Na2HPO4), sodium dihydrogen phosphate (NaH2PO4), sodium

3-(trimethylsilyl)-propionate-2,2,3,3-d4 (TSP), sodium azide (NaN3) and deuterium

oxide (D2O)and then analyzed using NMR spectroscopy. Data analysis was done

using multivariate analysis including principal component analysis (PCA) and

orthogonal partial least squarediscriminateanalysis (OPLS-DA) followed by

univariate and multivariate logistic regression to develop a model to identify the AD

biomarkers. The reproducibility was done using intraclass correlation coefficient

(ICC).

xxi

For plasma, the OPLS-DA model revealed 39 bins with VIP (variable

influence on projection) value more than 1 significantly discriminated AD from

social drinkers and controls. The sensitivity, specificity and accuracy of the model

were 64.29%, 98.17% and 91.24% respectively. For urine, the OPLS-DA model

revealed 59 bins with VIP value more than 1 significantly discriminated AD from

social drinkers and controls. The sensitivity, specificity and accuracy of the model

were 86.21%, 97.25% and94.93% respectively. In the univariate logistic regression

analysis of plasma, 9 peaks were significantly associated with AD with the p value ≤

0.1. In the multivariate logistic regression analysis, 4 peaks were significantly

associated with AD with area under the receiver operating coefficient (AUROC)

0.961. The sensitivity, specificity and accuracy of the model were 78.6%, 98.2% and

94.2% respectively. In the univariate logistic regression analysis of urine,30 peaks

were significantly associated with AD with the p value ≤ 0.1. The PCA revealed 5

PCs were significantly associated with AD. In the multivariate logistic regression

analysis, 2 PCs which consisted of 18 peaks were significantly associated with AD

with AUROC 0.909. The sensitivity, specificity and accuracy of the model were

65.5%,99.1%and92% respectively.

In this study, we have shown that the applied plasma and urine metabolomics

technique was able to differentiate between the alcohol-dependent and the social

drinkers and the controls. Nuclear magnetic resonance based metabolomics was also

able to identify novel biomarkers in plasma and urine which can be useful to

diagnose AD in the future.

1

CHAPTER 1

INTRODUCTION AND LITERATURE REVIEW

1.0 Background

1.1 The global burden of alcohol drinking

Alcohol is one of the most historical substances that has been misused by

man until today. Alcohol drinking is a ubiquitous public health problem. Its

detrimental effect does not only ravage the consumers but also their families and

societies. The World Health Organization (WHO) reported 2.5 million deathsevery

year due to the hazardous consumption of alcohol(IAS, 2013; WHO, 2011). The

United States reported 24,518 death cases due to alcohol consumption in 2009, of

which 15,183 of them died due to alcoholic liver cirrhosis(NCHS, 2009). Many

studies have shown that the harmful effects of alcohol consumption is the highest

compared to other illicit drugs(van Amsterdam & van den Brink, 2013). Besides its

harmful effect on vital organs such as liver and kidney(Lieber, 2003), alcohol also

has negative effects on cognitive perception, mood and behavior. Alcohol

intoxication usually leads to drastic and destructive behavior. This will create social

problems to the community. It also increases the incidence of accidents, injuries and

crimes(Couture, Brown, Tremblay et al., 2010; Vaaramo, Puljula, Tetri et al., 2012).

Alcohol consumption is estimated to be the third risk factor for disease

burden(Lieber, 2003).It was reported that there are some caseswhere alcohol

consumption causedimmune system impairment(Karavitis & Kovacs, 2011).

Therefore, chronic alcohol consumers usually have high susceptibility to acquire

fatal infections(Cheng & Currie, 2005; Romeo, Warnberg, & Marcos, 2010).

2

Moreover, it has a negative impact on the vascular system by causing hemorrhagic

stroke, arrhythmia and high blood pressure(McKee & Britton, 1998; J. Rehm, Room,

Graham et al., 2003).

1.2 Alcohol consumption in Malaysia

In Malaysia, according to the WHO report in 2005, alcohol consumption has

increased since 2001 with beer as the highest consumed type of alcohol (WHO,

2005).

A research that was conducted to investigate the effects of alcohol drinking and drug

abuse on fatal road accidents in urban areas of Kuala Lumpur, between year 2006

and year 2009, indicated that 23.3% of fatal drivers were positive to alcohol, 11%

positive to drug abuse and 2.3% were positive to both(Norlen Mohamed, Batcha,

Nurul Kharmila Abdullah et al., 2012). A recently published study, that aims to find

the etiologies of liver cirrhosis in Malaysia, revealed that alcohol is the main cause of

liver cirrhosis among Indian Malaysians, and the fourth cause of liver cirrhosis

among Chinese Malaysians(Qua & Goh, 2011). Such scrutiny sheds light on the

increasing harmful effects of chronic alcohol consumption in Malaysia. In fact, this

requires a thorough investigation of possible solutions that can early prevent the

devastating consequences of chronic consumption of alcohol.

1.3 Organ damage induced by alcohol consumption

A growing body of literature has asserted the harmful effects of alcohol

consumption to the vital organs. Although some studies advocated cardio protective

3

role of moderate alcohol consumption(Boto-Ordonez, Urpi-Sarda, Queipo-Ortuno et

al., 2013; Wallerath, Poleo, Li et al., 2003), other studies contradicted this by

indicating that the protective role is overvalued(Fillmore, Stockwell, Chikritzhs et

al., 2007; Plunk, Syed-Mohammed, Cavazos-Rehg et al., 2013; van Amsterdam &

van den Brink, 2013). Such contentious literature robustly defeat any suggested

beneficial effect of alcohol in the prevention of cardiovascular diseases.

In addition to cardiovascular diseases (CVDs), previous studies have found

that the average volume of alcohol consumption is positively correlated with liver

cirrhosis, depression, gastric ulcer and osteoporosis(Aguilera-Barreiro Mde, Rivera-

Marquez, Trujillo-Arriaga et al., 2013; Fini, Salamanna, Veronesi et al., 2012;

McKee & Britton, 1998; J. Rehm et al., 2003; van Amsterdam & van den Brink,

2013). It was concluded that alcohol consumption causes malabsorption, oesophageal

reflux and diarrhea(Bujanda, 2000). Besides,alcohol drinkingalso reduces fertility in

both males and females (van Amsterdam & van den Brink, 2013). The literature has

asserted that different patterns of alcohol consumption is a risk factor of several types

of cancer such as liver cancer, breast cancer, colon cancer, renal cell cancer and

oropharyngeal cancer(Baan, Straif, Grosse et al., 2007; Bagnardi, Rota, Botteri et al.,

2013; Bujanda, 2000; Radoi, Paget-Bailly, Cyr et al., 2013; J. Rehm et al., 2003). A

recent meta analysis that aims to investigate the association between light alcohol

drinking and different types of cancer has concluded that light drinking is associated

with oropharyngeal cancer, breast cancer and oesophageal cancer(Bagnardi et al.,

2013). In other words, chronic alcohol consumption can lead to injury and then to

subsequent failure of vital organs. This requires some effort to be made to encourage

the avoidance of alcohol consumption. Moreover, the early identification of people

4

who might be under the harmful effects of chronic alcohol consumption will help to

save their vital organs.

1.4 Alcohol metabolism in human

The alcohol is metabolized in the human body to acetaldehyde using three

enzymes. These enzymes are alcohol dehydrogenase (ADH), cytochrome P450

(CYP2E1) and catalase, which work by oxidizing ethanol to acetaldehyde. After that,

the acetaldehyde will be metabolized to acetic acid by acetaldehyde dehydrogenase

(ALDH). These two steps of metabolizing ethanol to acetic acid will increase the

ratio of NADH/NAD+ ratio (Zakhari, 2006).

Figure 1.1: Ethanol metabolism in human body

1.5 Alcohol-dependence (AD)

The National Health Committee, Wellington-New Zealand, in their

guidelines for recognizing, assessing and treating alcohol and cannabis abuse in

5

primary care, has adapted a definition for alcohol dependence (AD) based on the

American Psychiatric Association's classification system (DSM-IV), as the

following(BPAC, 2010; "Guidelines for Recognising, Assessing and Treating

Alcohol and Cannabis Abuse in Primary Care," 1999):

"A maladaptive pattern of alcohol use leading to clinically significant impairment or

distress".

This distress has many symptoms, such as a desire to drink more amounts of

alcohol to attain intoxication effect, spending a lot of time either consuming alcohol

or recovering from it, refraining fromsocial life activities, searching for events or

places that will include drinking alcohol, making ineffective efforts to cut down

alcohol consumption and having physical disorders after reducing the amount of

consumed alcohol such as tremors, anxiety, hallucination, nausea and

vomiting(BPAC, 2010; "Guidelines for Recognising, Assessing and Treating

Alcohol and Cannabis Abuse in Primary Care," 1999).

Unfortunately, the data shows that when the habit of drinking alcohol starts

earlier in age, the susceptibility to develop AD is higher(Grant & Dawson, 1997). It

could be inferred that when individuals become known as alcohol dependent, they

would already have gone through a long period of exposure to the detrimental effects

of alcohol consumption and hence will suffer failure of some of their vital organs. As

an early diagnosis of diseases will lead to an early start of treatment, early detection

of AD is crucial for preventing organs failure. For instance, the early detection of AD

might prevent the progression of liver injury to anirreversible phase.

AD can be treated by ALDH inhibitor such as disulfiram (Carroll, Nich, Ball

et al., 1998), or by opioid antagonist such as naltrexone (Volpicelli, Alterman,

6

Hayashida et al., 1992), or by topiramate(Johnson, Ait-Daoud, Bowden et al.) in a

standardized medication compliance management of AD.

1.5.1 Current diagnosticmethods of alcohol-dependence (AD)

Currently, AD questionnaires are the main clinical methods to diagnose the

disease in clinical practice. Some biomarkers can also be used to aid the diagnosis.

As biomarkers are usually produced by the body in response to disease or organ

injury(Kumar & Sarin), a group of biomarkers were deduced from studies which

investigated the harmful effects of alcohol consumption on body organs(Adias,

Egerton, & Erhabor, 2013; Freeman & Vrana, 2010; Litten, Bradley, & Moss, 2010).

Therefore, the biomarkers of organ injury, such as liver injury, are used as

biomarkers of AD as well(Adias et al., 2013).However, their specificity to AD is

reduced because they are biomarkers of organ injury regardless of the cause of

injury(Adias et al., 2013; Freeman & Vrana, 2010). These biomarkers may lead to a

confusionof diagnosis of AD with other diseases, such as non-alcoholic liver

cirrhosis(Litten et al., 2010).

1.5.1.1 AD questionnaires

Several questionnaires have been developed to indicate alcohol dependence

diagnosis. The commonly used questionnaires are the alcohol use disorders

identification test (AUDIT), the short version of AUDIT (AUDIT-C) and Michigan

alcoholism screening test (MAST)(BPAC, 2010). These questionnaires consist of a

set of questions that can differentiate between harmful and dependent alcohol

consumption, and predict the leading risk factors of some diseases associated with

alcohol consumption.

7

The main drawback of AD diagnostic questionnaires is their subjective nature

of measurement. They may not provide an actual estimation of an individuals'

alcohol consumption due to the denial and under reporting of the regular consumed

amount of alcohol. Moreover, there are limitations in the validations of these

questionnaires due to the lack of standard questionnaires or definite diagnostic tool to

be used as validation reference(Kroke, Klipstein-Grobusch, Hoffmann et al., 2001).

1.5.1.2 Biomarkers

Biomarkers are biological indicators of a specific medical condition or

disease which can be tested or measured by using lab tools. The National Institute of

Health (NIH)Biomarkers Definitions Working Group had defined a biomarker as " a

characteristic that is objectively measured and evaluated as an indicator of normal

biological processes, pathogenic processes or pharmacologic responses to a

therapeutic intervention" (NIH Biomarkers Definitions Working Group,

2001)(Freeman & Vrana, 2010). For example, the increase in serum creatinine level

and consequently the decrease in creatinine clearance are used as biomarkers to

diagnose and monitor renal diseases(Duncan, Heathcote, Djurdjev et al., 2001;

Myers, Miller, Coresh et al., 2006).

The second method to diagnose AD is by testing AD biomarkers in biofluids.

In fact, studies that were conducted to investigate the mechanisms of alcohol induced

organ damage have found several mechanisms/pathophysiology of injuries involving

certain biomarkers. For example, the rise of liver enzymes such as Alanine

Aminotransferase (ALT), Aspartate aminotransferase (AST) and Gamma

Glutamyltransferase (GGT) in AD individuals was obviously an indicator of alcohol

induced liver injury(Adias et al., 2013). These differences in the levels of AST and

8

GGT in AD individuals compared to their abstainers had been found to be significant

in several studies(Adias et al., 2013; Quaye, Nyame, Dodoo et al., 1992).The altered

level of inflammatory mediators such as (cytokines) due to alcohol consumption had

been correlated with the harmful effects of alcohol on bone, lung, liver and other

tissues(Birkedal-Hansen, 1993; Fini et al., 2012; N., Freeman, & Vrana, 2011).

Furthermore, as coagulating factors are synthesized in the liver, a study has

concluded that prothrombin time (PT) and activated partial thromboplastin time

(APTT) are significantly elevated in AD individuals (Adias et al., 2013).

Therefore, today it is a common practice to use blood and urine biomarkers of

organ damage to diagnose AD. There are many biomarkers that have been

investigated in research to diagnose AD such as Gamma glutamyl transferase (GGT),

Mean Corpuscular Volume (MCV), Carbohydrate-Deficient Transferrin (CDT),

Phosphatidylethanol (PEth), Ethyl Glucuronide (EtG) and Ethyl Sulfate

(EtS)(Freeman & Vrana, 2010; George G. Harrigan, Maguire, & Boros, 2008; John

C. Lindon, Nicholson, & Holmes, 2007 ; Litten et al., 2010; Rinck, Frieling, Freitag

et al., 2007).

Gamma glutamyl transferase (GGT) is available in cell membranes of the

tissues of some organs such as liver, kidney, spleen, pancreas and heart(Litten et al.,

2010). Chronic alcohol consumption results in inflammation and necrosis of these

cells, consequently this might increase the leakage of GGT from the destroyed cells

(especially hepato-cells) thereby leading to an elevation in serum GGT(Litten et al.,

2010; Tavakoli, Hull, & Michael Okasinski, 2011).GGT has a long window of

assessment, it remains elevated for two to three weeks after alcohol cessation, and it

takes two weeks to elevate after the relapsing of heavy drinking(BPAC, 2010).It has

20% sensitivity and 65.2% specificity (Larkman, 2013).

9

The Mean Corpuscular Volume (MCV)is a measurement of the size of red

blood cells. It is an indicator of the chronic use of alcohol rather than the acute

intake(Tavakoli et al., 2011), as chronic use increases the size of red blood

cells(BPAC, 2010).The level of MCV can remain high for several months after

cessation (Litten et al., 2010). Compared to GGT,the sensitivity of MCV in women is

greater than that in men(Conigrave, Davies, Haber et al., 2003).

Carbohydrate-Deficient Transferrin (CDT) is the widely used AD biomarker

in clinical practice. Transferrin is glycoprotein synthesized and released by the liver,

and transports iron throughout the body. Regular high alcohol intake leads to a

decrease in the number of carbohydrate residues (sialic acid) attached to transferrin,

thereby increasing carbohydrate deficient sites(BPAC, 2010). Serum CDT elevates

with regular and heavy drinking(60-80 g per day) and usually returns to normal after

two to three weeks of cessation(Stibler, 1991). Because of the improvement in CDT

measurement methods, the sensitivity and specificity of its testing have increased.

Therefore, the Food and Drug Administration (FDA) affirmed CDT as a biomarker

of heavy alcohol consumption(Litten et al., 2010).

Phosphatidylethanol (PEth) is a phospholipid formed by phospholipase D

enzyme in the presence of ethanol. It remains elevated in the blood for one to two

weeks after cessation of moderate to heavy alcohol intake(Stewart, Reuben,

Brzezinski et al., 2009). The advantage of PEth is that it is more sensitive than other

biomarkers as it is not affected by the liver state(Stewart et al., 2009). Although other

biomarkers can be increased by non-alcohol liver diseases, PEth can be used to

monitor alcohol consumption with hepatic disease(Litten et al., 2010).

10

Ethyl Glucuronide (EtG) is a direct metabolite of ethanol, as ethanol is

conjugated in hepatic endoplasmic reticulum with glucuronic acid through

glucuronosyltransferase enzyme into EtG. It can be measured in blood, urine and

hair. Urine EtG is usually used rather than serum EtG due to the short half-life of

serum EtG which is between 14-20 hours. Whereas, urine EtG can be detected up to

several days. Currently, there are developed methods to measure EtG in hair and this

will help to detect EtGfor several months after alcohol cessation(Litten et al., 2010).

Ethyl Sulfate (EtS) is a metabolite of alcohol. It is yielded from the sulfate

conjugation of alcohol by sulfo-transferase enzyme. It was discovered that there is a

positive correlation between EtG and EtS in urine samples(Litten et al., 2010).

1.5.1.2.1 Limitations of thecurrentAD biomarkers

Although current AD biomarkers are being used to assist in making AD

diagnosis, there are some limitations that hinder their uses. The drawbacks are

mainly due to their low specificity and sensitivity which indeed lead to false

diagnosis.

The AST and ALT are not specific to AD, they are usually used to screen

liver damage regardless of the cause(Adias et al., 2013; Litten et al., 2010).The GGT

is also not specific to AD. It can be elevated in non-alcoholic liver diseases, smoking,

obesity, diabetes mellitus, age, nutritional factors and metabolic disorders(Adias et

al., 2013; Litten et al., 2010). Moreover, GGT can be elevated by some medications

such as barbiturates, anticonvulsants and anticoagulants(Litten et al., 2010). In fact,

the low sensitivity and specificity will hinder the use of GGT, AST and ALT in the

discrimination between alcoholic and non-alcoholic liver diseases.

11

The MCV is elevated in both folate and vitamin B12 deficiencies,

hypothyroidism, non-alcoholic liver diseases, hemolysis, and bleeding disorders. It

can also be elevated in patients on treatment with medications that can induce bone

marrow disorders and toxicity(Litten et al., 2010).

Though PEth has the advantage of being a specific biomarker of alcohol

consumption and not affected by liver condition, it is limited by its lower

differentiation ability between moderate and heavy drinkers(Stewart et al., 2009).

Similarly, the EtG has its limitations as well. For example, if urine sample contains

yeast, this will convert urine glucose to alcohol and then to EtG.This may lead to

false positive results,especially of diabetic patients who have high levels of glucose

in urine(Litten et al., 2010).Moreover, washing the hair may remove EtG from hair

shaft(Tavakoli et al., 2011).The EtG's window of assessment isbetween one to two

days after alcohol cessation which is too short a time to detect AD.

The CDT sensitivity is similar to GGT, but with higher specificity. A recent

study to evaluate and to compare the range of AD biomarkers of a group of heavy

drinkers in Russia concluded that CDT might be the best biomarkers with 67%

sensitivity and 71% specificity to detect a daily average alcohol consumption of 40g

and above(McDonald, Borinskya, Kiryanov et al., 2013). However, the use of CDT

as AD biomarker is hindered by the possibilities of false positives due to rare genetic

transferrin variants, chronic end-stage liver disease, smoking, body weight, female

gender, primary biliary cirrhosis and hepatocarcinoma(Litten et al., 2010).

In an approach to increase sensitivity and specificity of AD biomarkers, some

combinations had been examined. The best suggested combination is CDT, GGT and

MCV(Larkman, 2013; Litten et al., 2010). This combination might help in males and

12

females, heavy drinkers with or without liver diseases(Litten et al., 2010).Another

combination is CDT and MCV which has been shown to improve sensitivity and to

perform better than either one of the biomarkers alone. However, the risk of false

diagnosis cannot be excluded(Tavakoli et al., 2011).

Nevertheless, there were some reports which stated that the CDT, GGT and

MCV are not reliable enough biomarkers of AD,the current practice uses them to aid

questionnaires(Larkman, 2013).Therefore, a combination of biomarkers was

suggested to limit but not rule out the risk of false diagnosis(Larkman, 2013). In

general, the questionnaires and current AD biomarkers are not the perfect methods to

diagnose AD accurately. Therefore, it is reasonable to argue that the finding is a

more objective measurement of novel AD biomarkers that would be of an advantage.

The diagnosis of AD patients will help to reduce AD patients' organ failure and

social problems. It has been suggested that the use of any of the systems biology

disciplines, particularly metabolomics, could eventually prove valuable in this

area(Kumar & Sarin).

1.6 Systems Biology

System biology is a biological approach that aims to investigate and to

explore the complexity of molecular perturbation by the comprehensive integration

of different bio-databases of molecular, genetic and metabolic networks, and the

individual interaction between the components of each network(Breitling, 2010;

Kuster, Merkus, van der Velden et al., 2011; G. Louridas & Lourida, 2012; G. E.

Louridas, Kanonidis, & Lourida, 2010). This can give an in-depth understanding of

the medical condition and hence improvesdrug discovery and personalization of

13

therapy(G. Louridas & Lourida, 2012). In fact, the power of system biology is based

on the concept that although the clinical phenotype of a molecular disturbance is not

obvious, the consequent compensatory adaptation due to this disturbance will be

reflected in the transcriptome, proteome or metabolome(Kuster et al., 2011).

1.6.1 Disciplines of Systems Biology

Systems biology combines a group of (-omics) disciplines, particularly the

main majoromics such as genomics, transcriptomics, proteomics and

metabolomics.While genomics focuses on the genome sequencing,

transcriptomicsstudies the transcription and expression of genes sets using gene

expression microarrays or RNA sequencing (Cappola & Margulies, 2011; Piran, Liu,

Morales et al., 2012). These expressions will yield proteins which can be measured

by gel electrophoresis or mass spectrometry (MS). The biochemical modifications to

proteins can reflect a state of specific medical conditions. Lastly, the metabolic

process in cells and organs ends with the production of metabolites which constitute

the metabolome or metabolic profile.Metabolomics aim to discover and measure the

changes in metabolome. Each one of the systems biology disciplines has been found

to give fingerprints of prognostic value in diagnosing diseases.However,

metabolomics is the ultimate screen that reflects other omics in addition to

environmental factors. The use of systems biology disciplines to diagnose

complicated medical conditions such as cancer, cardiovascular diseases and liver

disease, has the advantage of being less invasive and fast(Cappola & Margulies,

2011; Piran et al., 2012).Figure1.1 shows how the systems biology disciplines help in

the understanding of the pathophysiology of diseases and the pathways behind

specific phenotypic manifestation of the disease. The figure shows that metabolomics

14

is the ultimate internal picture of systems biology which can reflect the preceding

disciplines.

Figure 1.1: The role of systems' biology disciplines (omics) in disease diagnosis, where metabolomics represents the ultimate biomarker for detection of the disease before the phenotypic manifestations (adapted from(Piran et al., 2012)).

1.7 Metabolomics

Metabolomics (ormetabonomics in some literature) is one of the systems'

biology disciplines which comes in a chaincircle after genomics, transcriptomics and

proteomics. It deals with the identification and quantification of small molecular

metabolites in the metabolic profile (metabolome) of the living organism(Corona,

Rizzolio, Giordano et al., 2012; G. Louridas & Lourida, 2012). The metabolome of

each organism is highly affected either byinternal or external environmental factors.

Therefore, the variation in metabolomemight be associated witha particular

phenotype, specific nutrition, drugsand diseases. This variation can be considered as

the metabolic finger print or surrogate biomarkers of medical condition or disease.

Metabolomicsalso helps in the ability to understand the pathways that lead to

changes in the levels of body metabolites due to disease, drug and/or environmental

Genomics

Genetic Predisposition

Transcriptomics

Biomarkers

Disease Phenotypic

Manifestation

Metabolomics Biomarkers

Proteomics Biomarkers

Env

irom

enta

l fac

tors

Env

irom

enta

l fac

tors

Env

irom

enta

l fac

tors

15

effect(George G. Harrigan et al., 2008).The reason for the use of metabolomics in the

exploration of medical conditionsin recent medical approach is because it doesnot

only reflect the variation in genetics, transcriptomics and proteomics which might be

associated with the condition, but also the environmental factors associated with the

condition (Gutiu, Andries, Mircioiu et al., 2010). Moreover, it can give a

predictionof body response to these internal and external factors. This can help in

drug discovery and personalization of therapy.Recent studies have used

metabolomics analysis to identify novel biomarkers of diseases such as asthma,

COPD, cancer, diabetes mellitus, cystic fibrosis and metabolic disorders (Hocquette,

2005; Hunt & John, 2007; Montuschi, Paris, & Melck, 2009; Robroeks, Van Berkel,

Dallinga et al., 2010).

The metabolomics analysis involves the identification and measurement of

small molecularcompounds inbiological samplessuch as blood, urine, cerebrospinal

fluids CSF, breath, seminal fluids and tissues. These small molecular compounds are

metabolites. Thetechniqueinvolveschemometric studies of the spectroscopic data of

instruments such as nuclear magnetic resonance (NMR)spectroscopy and mass

spectrometer (MS)(John C Lindon, Nicholson, & Holmes, 2011).

There are two terms that are commonly used to describe the study of body

metabolites which are metabolomics and metabonomics. Metabonomics is defined as

the study of metabolic response of living organism to biological or genetic variation

(Nicholson & Lindon, 2008), while metabolomics is defined as the exploration of all

metabolites inside the living organism(Fiehn, 2001; R. Goodacre, Vaidyanathan,

Dunn et al., 2004). The distinction between these two terms; metabolomics and

metabonomicsremains controversial and both terms are being used interchangeably

in the literature to refer to the same discipline.In this context we are going to use the

16

term metabolomics, however, we might also refer to some literature which used the

term metabonomics instead for our literature review.

1.8 Application of metabolomics in disease diagnosis

As the pathophysiological changes in the body are usually reflected on the

metabolome, it can be inferred that different medical conditions can have their

distinct metabolic finger print (metabotype)on the metabolome. The metabotype is a

group of certain metabolites that, eitherby their presence or absence, increased or

decreased concentration,are distinctive for a specific clinical status or

disease(Semmar, 2012).Researchers have investigated the use of metabolomics

techniques to explore different medical conditions, in particular the conditions which

have no definitediagnosis or their clear-cut diagnosis which needfurther invasive

procedures.

1.8.1 Application of metabolomics techniques to identify novel-biomarkers of

diseases

The aim of metabolomics studies is to find a metabotypewhich has diagnostic

or classifying value among patients andto exploreand measure the metaboliteswhich

will further assistin the diagnosis or classification(John C. Lindon et al., 2007 ).

Many metabolomics studies have been conducted on biological fluid samples

to identify biomarkers of diseases such as liver disease, diabetes, asthma, cancer and

critical illnesses(Serkova, Standiford, & Stringer, 2011; Wang, Zhang, Han et al.,

2012). These studies have identified metabotypes which are able to

17

discriminatepatients from their controls. Table 1.1presents examples of

metabolomics studies to diagnose diseases.

Some of the metabolomics studies aim to investigate metabolic changes

consequent to the exposure to environmental factors or toxin, such as smoking and

other harmful substances. In an approach to explore the effects of smoking and

smoking cessation, researchers have used metabolomics techniques to quantify 140

metabolites in fasting serum of three groups, namely current smokers, non-smokers

and quitters (who have quitted during the follow up period of the study) in a

longitudinal analysis (Xu, Holzapfel, Dong et al., 2013). It was discovered that 21

smoking related metabolites were significantly different from those of current

smokers and non-smokers. Interestingly, the study discovered that 19 out of the 21

metabolites were reversible in quitters (Xu et al., 2013). In another study aimed to

study the systemic toxic effect of welding fumes on humans, Wei and colleagues

used liquid and gas chromatography-MS to investigate the plasma metabolome of

boilermakers pre-welding and post-welding fumes exposure in a two stage-study

(Wei, Wang, Chang et al., 2013). The first stage (stage one) of the study was

conducted in 2011 on 11 boilermakers. The second stage (stage two) was conducted

in 2012 on 8 boilermakers, and five of them participated in stage one in addition to

three new recruited boilermakers. The results showed that the high exposure to high

metal welding fumes causes a decrease in the level of unsaturated fatty acids (Wei et

al., 2013).

The aforementioned metabolomics studies are part of continually growing

body of evidences on the preeminent role that metabolomics can play not only in

diseases diagnosis, but also in the pathogenic understanding of diseases and variable

18

medical conditions. This will help to monitor, guide and evaluate the current therapy,

and help to find new drug target for future drugs.

Hitherto, to our knowledge, there were no metabolomics studies that had been

conducted in humans to identify biomarkers of chronic use of alcohol or AD.

Therefore, we propose the use of metabolomics in this research to find novel

biomarkers that can discriminate AD patients from their matched controls and their

matched social drinkers.

19

Table 1.1: Examples of metabolomics studies to diagnose diseases

Study Disease Analytical Method Specimen Findings

Ibrahim et al.,2011 (Ibrahim, Basanta,

Cadden et al., 2011) Asthma GC-MS Breath

Development of a discriminatory model which classifies asthma patients with accuracy of 86%

Basanta et al.,2012 (Basanta, Ibrahim,

Dockry et al., 2012) COPD GC-MS Breath

Development of a discriminatory model which discriminate COPD patients from their healthy controls with 85% sensitivity and 50% specificity

Schicho et al.,2012 (Schicho,

Shaykhutdinov, Ngo et al., 2012)

IBD 1HNMR Serum,

Plasma and Urine

Characterization of 44 serum, 37 plasma and 71 urine metabolites to differentiate between diseased and non-diseased individuals

Motsinger-Reif et al., 2013 (Motsinger-Reif, Zhu, Kling et al., 2013)

Al-zheimer with

Dementia

LC-ECA GC-TOF MS

CSF Identified biomarkers which discriminate Al-zheimer patients from their healthy controls

Lewitt et al., 2013

(Lewitt, Li, Lu et al., 2013)

Parkinson’s disease (PD)

UPLC and GC- MS CSF Identification of 19 compounds which were able to discriminate PD patients from similar age healthy controls with a false discovery level of 20%

GC-TOF MS: gas chromatography-time of flight mass spectroscopy, CSF: cerebrospinal fluid, UPLC: ultra performance liquid chromatography, IBD: Inflammatory bowel disease, COPD: Chronic obstructive pulmonary disease, LC-ECA: liquid chromatography-electrochemical coulometric array

20

1.8.2 Using metabolomics to investigate alcohol consumption

There are metabolomics studies that investigated the effect of alcohol

consumption by analysing biological samples of animals on alcohol-containing diets.

Further, researchers started to explore this in humans.

1.8.3 Using metabolomics to investigate alcohol consumption in animal models

There are some studies that used the metabolomics approach to study the

pathogenesis of alcohol consumption in animal models and to find novel biomarkers

of alcohol consumption. Most of these studies were oriented by expected

consequences of alcohol's detrimental effect such as liver injury or alcoholic liver

disease (Bradford, O'Connell, Han et al., 2008; Fernando, Kondraganti, Bhopale et

al., 2010; Loftus, Barnes, Ashton et al., 2011).

Bradford and colleagues used 1H-NMR and MS metabolomics techniques to

evaluate the metabolic profile of urine and liver extract samples of mice with alcohol

induced liver injury (Bradford et al., 2008). The mice were divided into two groups,

one was fed with isocaloric (control group) and the other group was fed with alcohol

containing liquid diet (alcohol group) of which the steatohepatitis was confirmed by

5-fold increase of serum ALT, 6-fold increase in liver injury score and the increase

of lipid peroxidation in the liver. The 1H-NMR principal component analysis of both

urine and liver extract showed obvious discrimination between the two groups. For

instance, the lactate was high in both liver and urine of those mice in the alcohol

group. N-oleoylethanolamine metabolite was found to be elevated as well. Both

lactate and N-oleoylethanolamine indicated hypoxic injury of the liver. Tyrosine was

found to be elevated which might reflect an alteration in metabolism due to alcohol.

21

Moreover, there was a decrease in the excretion of taurine (glutathione metabolite) in

the urine of those mice in the alcohol group. Additionally, it was found that there was

an increase in prostacycline inhibitor 7,10,13,16-docosatetraenoic acid which is vital

in the regulation of platelets formation (Bradford et al., 2008).

In another study, by using LC-MS metabolomics technique, Loftus and

colleagues explored the metabolomics changes in non-polar metabolites of rodents'

livers due to alcohol consumption (Loftus et al., 2011). The study was conducted on

rats and mice fed with intragastric alcohol feeding model. The analysis of liver

derived samples revealed a significant increase of fatty acyls, fatty acid ethyl esters,

in addition to octadecatrienoic acid and eicosapentaenoic acid metabolites (Loftus et

al., 2011). Similarly, but by targeting the lipidomic profile, Fernando and colleagues

extracted lipids from the plasma and liver of rats fed with alcohol diets and their

controls (Fernando et al., 2010).It was concluded from both 1H-NMR and 31P-NMR

data analysis that a significant alteration in lipid metabolism was induced by alcohol

consumption.

To the best of our literature review we have found that most of the

metabolomics studies which assessed alcohol consumption in animal models,

focused on specific organ injury, particularly liver injury. None of these studies

aimed to find the biomarkers of chronic use of alcohol or AD.

1.8.4 Using metabolomics to investigate alcohol consumption in humans

To date, there are few metabolomics studies that have been conducted to

study the effects of alcohol consumption on human metabolome. Jaremek and

colleagues investigated the effects of alcohol consumption on human serum

metabolome(Jaremek, Yu, Mangino et al., 2013). Researchers compared the serum

22

metabolome of two groups, light (LD) and moderate to heavy drinkers (MHD). The

results showed that 40 identified metabolites in males and 18 in females differed

significantly in concentration between these two groups. Out of these metabolites, 10

in males and 5 in females were specific metabolites to discriminate LD from MHD.

The investigators concluded that alcohol consumption mostly affect metabolic profile

classes of diacylphosphatidylcholines, lysophosphatidylcholines, ether lipids and

sphingolipids. These results indicated that the stimulatory effect of alcohol

consumption on acid sphingomyelinase (ASM) activity causes the accumulation of

ceramide and a decrease of sphingomyelins(Jaremek et al., 2013). However, the

study did not explore the metabolic variation associated with chronic use of alcohol

or AD.Additionally, it did not investigate urine metabolomics which is non invasive

compared to blood.

In another study, the researchers used 1H-NMR metabolomics technique to

find serum metabolic finger print that discriminate alcoholic liver cirrhosis from

hepatitis B virus liver cirrhosis (Qi, Tu, Ouyang et al., 2012). The investigators found

that five metabolites were able to distinguish between the two different types of

cirrhosis. Similar to the studies mentioned earlier, the study did not explore urine

metabolic profile and did not look for metabolic finger print of chronic use of

alcohol.

1.9 Application of metabolomics to find novel biomarkers of ADin blood and

urine samples

As metabolomics studies aim to find a specific metabolic fingerprint

(metabotype) that is associated with a particular medical condition or

23

disease,samplinga suitable biological sample is required.Blood and urine samples are

usually considered stable and less invasive biological sampleswhere urine is

leastinvasive and easy to get in abundance as it is the waste product (Decramer, de

Peredo, Breuil et al., 2008; Down, 2010; Griffiths, 2008; Vaidyanathan, Harrigan, &

Goodacre, 2005; Want, Wilson, Gika et al., 2010). Therefore, researchers

preferplasma, serum and urine in metabolomics studies.

1.10 Using 1H-NMR spectroscopicinstrument in metabolomics

The proton nuclear magnetic resonance 1H-NMR is one of the main analytical

tools that isbeing widely used in metabolomics, in addition to mass spectroscopy

(MS) and fourier transform infrared spectroscopy (FT-IR) (Basanta et al., 2012;

Lloyd, William Allwood, Winder et al., 2011; Schicho et al., 2012). It is being used

increasingly in metabolomics recently because it has the advantage of quickly

revealing the metabolic profile of the biological sample depending on the magnetic

properties of the widely spread hydrogen atoms in the chemical structure of the

metabolites(Dunn, Bailey, & Johnson, 2005). The concept that1H-NMR

analysisprocesses rely on every compound or metabolite contains hydrogen atoms in

different chemical structure forms that will give specific and different peaks in the

NMR spectra when the compound enters the magnetic field. These peaks are

characteristic for each compound(Dunn et al., 2005).This, with the help of a

chemometricsoftware,will help to speculate the chemical structure of the unknown

metabolites in the biological samples.Although the sensitivity of 1H-NMR is low

when compared to MS (Silva Elipe, 2003), the1H-NMR analysis is less complex as

the sample does not need a prior derivatization, extraction and separation as required

24

by MS (Clayton, Lindon, Cloarec et al., 2006). In addition, 1H-NMR acts as an

independent instrument compared to MS which requires a separate instrument like

GC or LC(Dunn et al., 2005). The 1H-NMR is able to analyze hundreds of samples

per day and the sample can be reused for further analysis (Nicholson & Lindon,

2008; Shulaev, 2006). Therefore, the1H-NMR is considered cheap and non

destructivewhen compared to MS. Figure 1.2 illustrates a Google scholar based

literature search for publications containing (metabolomics & NMR) and

(metabonomics & NMR). The search showed an increase in the number of

publications from a total of 67 articles in year 2000 to 5870 articles in 2014. This

reflects the great interest in using NMR in metabolomics studies due to the simplicity

of sample preparation and the rapidyield of results.

Figure1.2: Literature search at Google scholar for publications of (metabolomics + NMR) and (Metabonomics + NMR)

0

1000

2000

3000

4000

5000

6000

7000

Metabonomics + NMR

Metabolomics + NMR